Original Article
In Vitro Recovery of Sufentanil, Midazolam, Propofol,
and Methylprednisolone in Pediatric Cardiopulmonary
Bypass Systems
Annewil van Saet, MD
*
,1, Gerdien A. Zeilmaker-Roest, MD
y,
Marloes P.J. van Hoeven, MSc
z, Birgit C.P. Koch, PhD
x,
Joost van Rosmalen, PhD
{, Martina Kinzig, PhD
k,
Fritz S
€orgel, BSc, PhD
k,#, Enno D. Wildschut, MD, PhD
y,
Robert J. Stolker, MD, PhD
*
, Dick Tibboel, MD, PhD
y,
Ad J.J.C. Bogers, MD, PhD
z*Department of Anesthesiology, Erasmus Medical Center, Rotterdam, The Netherlands
yDepartment of Intensive Care and Pediatric Surgery, Erasmus Medical Center, Rotterdam, The Netherlands zDepartment of Cardiothoracic Surgery, Erasmus Medical Center, Rotterdam, The Netherlands
xDepartment of Pharmacology, Erasmus Medical Center, Rotterdam, The Netherlands {Department of Biostatistics, Erasmus Medical Center, Rotterdam, The Netherlands kInstitute for Biomedical and Pharmaceutical Research, N€urnberg-Heroldsberg, Germany
#
Institute of Pharmacology, University Duisburg-Essen, Essen, Germany
Objectives: To evaluate in vitro drug recovery in cardiopulmonary bypass (CPB) systems used for pediatric cardiac surgery. Design: Observational in vitro study.
Setting: Single-center university hospital.
Participants: In vitro CPB systems used for pediatric cardiac surgery.
Interventions: Three full neonatal, infant, and pediatric CPB systems were primed according to hospital protocol and kept running for 6 hours. Midazolam, propofol, sufentanil, and methylprednisolone were added to the venous side of the systems in doses commonly used for induction of general anesthesia. Blood samples were taken from the postoxygenator side of the circuit immediately after injection of the drugs and after 2, 5, 7, 10, 30, 60, 180, and 300 minutes.
Measurements and Main Results: Linear mixed model analyses were performed to assess the relationship between log-transformed drug concen-tration (dependent variable) and type of CPB system and sample time point (independent variables). The mean percentage of drug recovery after 60 and 180 minutes compared with T1 was 41.7% (95% confidence interval [CI] 35.9-47.4) and 23.0% (95% CI 9.2-36.8) for sufentanil, 87.3% (95% CI 64.9-109.7) and 82.0% (95% CI 64.6-99.4) for midazolam, 41.3% (95% CI 15.5-67.2) and 25.0% (95% CI 4.7-45.3) for propofol, and 119.3% (95% CI 101.89-136.78) and 162.0% (95% CI 114.09-209.91) for methylprednisolone, respectively.
Conclusions: The present in vitro experiment with neonatal, infant, and pediatric CPB systems shows a variable recovery of routinely used drugs with significant differences between drugs, but not between system categories (with the exception of propofol). The decreased recovery of mainly sufentanil and propofol could lead to suboptimal dosing of patients during cardiac surgery with CPB.
Ó 2019 Elsevier Inc. All rights reserved.
Key Words: cardiopulmonary bypass; in vitro; midazolam; sufentanil; propofol; methylprednisolone
Cardiopulmonary bypass systems were made available free of charge by Terumo Europe NV, Leuven, Belgium, and Sorin Group, Mirandola, Italy.
1
Address reprint requests to Annewil van Saet, MD, Dr. Molewaterplein 40, Room Na1717, 3015 GD Rotterdam, The Netherlands. E-mail address:a.vansaet@erasmusmc.nl(A. van Saet).
https://doi.org/10.1053/j.jvca.2019.08.029
1053-0770/Ó 2019 Elsevier Inc. All rights reserved.
ARTICLE IN PRESS
Journal of Cardiothoracic and Vascular Anesthesia 000 (2019) 19
Contents lists available atScienceDirect
Journal of Cardiothoracic and Vascular Anesthesia
CARDIOPULMONARY BYPASS (CPB) is necessary to facilitate most cardiac surgery in children. The effect of CPB on in vivo drug concentrations in patients can be profound and is attributed to hemodilution, altered hemodynamic status, hypothermia, systemic inflammation, changes in acid-base sta-tus, exclusion of the lungs from the circulation, and hemofiltra-tion.1Furthermore, the plastic components of the CPB system themselves have been shown to absorb drugs.2-7
In the authors’ institution, multiple experiments have been performed to determine in vitro drug recovery in extracorpo-real membrane oxygenation (ECMO) systems.8,9Drug recov-ery has been defined as the concentration of drug present in the priming fluid after a certain amount of time has passed since addition of the drug to the CPB system.8,9Previous pub-lications have used the term absorption to indicate the decrease in concentration of drug in the priming fluid. In the authors’ institution, recovery has been deemed a more precise definition because not all drug is actually absorbed by the system compo-nents. Drugs also are subject to spontaneous degradation, for example, providing an altogether different reason for a decrease in drug concentration than absorption of drug to com-ponents of the CPB system.8There is a lack of data in the liter-ature concerning pediatric CPB systems. As part of the authors’ CPB PHARM study, which aims to measure and model drug concentrations during CPB for pediatric cardiac surgery (registered at the Netherlands Trial Register [NTR3579]), the in vitro experiments described herein were undertaken. The ultimate goal is to incorporate these data into in vivo population pharmacologic models.
Methods
The present study was conducted at the Department of Cardio-thoracic Surgery of a tertiary teaching hospital. No human partic-ipants were involved in the study, so the need for medical ethical review board approval was waived according to Dutch law.
Soon to be expired CPB systems were made available free of charge by Terumo Europe NV, Leuven, Belgium, and Sorin Group, Mirandola, Italy. This research did not receive any spe-cific grant from funding agencies in the public, commercial, or not-for-profit sectors. There was no role for Terumo Europe NV or Sorin Group in the design of the study, collection, anal-ysis, and interpretation of data; writing of the report; or the decision to submit the report for publication.
CPB Systems
Table 1shows the composition of the different CPB systems
used. All systems contained a hollow-fiber membrane oxygena-tor with a polymethylpentene membrane. For the neonatal and pediatric systems an arterial filter was integrated in the oxygena-tor, and for the infant system a stand-alone arterial filter was used. Silicone and polyvinylchloride (PVC) tubing with differ-ent lengths and diameters were used in the neonatal and infant roller-pump systems. In the pediatric system a centrifugal pump was used, and the silicone tubing was discarded. Tubing was made continuous via a one to-one fourth or a one
fourth-to-three eighths connection piece. A venous reservoir completed the systems. Terumo components of the systems were coated with X-coating (poly[2-methoxyethylacrylate]), which is a non-heparin biocompatible polymer with hydrophilic and hydropho-bic properties. Sorin components of the systems were coated with P.h.i.s.i.o. (Sorin) coating, which is a nonheparin, biomi-metic layer consisting of a phosphorylcholine polymer.
All systems were placed on a conventional mast-mounted, remote pump head console (St€ocker S5 Perfusion System; Sorin Group) with a specific pediatric configuration.
Three full systems were assembled for each category (neo-natal, infant, and pediatric) and primed according to hospital-based protocol (Table 2). Priming fluid contained fresh frozen plasma and Gelofusine (B. Braun, Melsungen, Germany). Red blood cells were added to the priming to achieve a hematocrit of 28%. Recently expired red blood cells and fresh frozen plasma obtained from the authors’ local blood bank were used for priming. The priming fluid was completed with human albumin (Sanquin Plasma Products BV, Amsterdam, The Netherlands) and 2 to 5 mL sodium bicarbonate 8.4% (Frese-nius Kabi Nederland BV, Zeist, The Netherlands). Heparin was added to the system according to hospital protocol to pre-vent clotting.
The CPB systems were kept running for 6 hours. This is the maximum runtime with guaranteed quality by the man-ufacturers. The temperature was maintained at 36˚C., pCO2, and pH were measured with an iStat handheld device (Abbot BV, Hoofddorp, The Netherlands) and main-tained within physiological ranges by titration of sweep gas flow, gas composition, and addition of sodium bicar-bonate 8.4% if needed.
A flow rate of 0.5 L/min was maintained for the neonatal circuits, 1.5 L/min for the infant circuits, and 3 L/min for the pediatric circuits. Postmembrane pressures were kept at 100 mmHg by adapting the resistance using the venous clamp. Drug Administration
For the neonatal system a standardized body weight of 5 kg was used for drug amount calculations, 15 kg for the infant system, and 30 kg for the pediatric system were used. Drugs were added to the venous reservoir via a manifold sample port in a dose that normally would be used for the induction of gen-eral anesthesia according to the following authors’ institution’s guidelines: midazolam (1 mg/mL; Actavis Group PTC ehf, Hafnarfj€ordur, Iceland) 0.2 mg/kg; propofol (10 mg/mL; Fre-senius Kabi Nederland BV) 2 mg/kg; sufentanil (50
m
g/mL; Hameln Pharma Plus GmbH, Hameln, Germany) 2m
g/kg; and methylprednisolone (100 mg/mL; Pfizer BV, Capelle a/d IJs-sel, The Netherlands) 30 mg/kg. Drugs were injected in the same order for all systems. Between administration of each drug and after administration of the last drug, the sample port was flushed with 2 mL of 0.9% saline solution to prevent crys-tallization or pooling of drug. Midazolam, propofol, sufentanil, and methylprednisolone were used because these drugs are commonly used for pediatric cardiac anesthesia in the authors’ institution.ARTICLE IN PRESS
Table 1 CPB Systems
Oxygenator Reservoir Arterial filter Venous Filter Cardiotomy
Defoaming Sponge
Silicone Tubing PVC Tubing Priming Volume Neonatal roller Capiox FX05 (Terumo
Europe NV, Leuven, Belgium) Hollow fiber Polycarbonate housing, polypropylene membrane 0.5 m2 Priming volume 43 mL X-coating
Open, hard shell polycarbonate Minimum capacity 15 mL Maximum capacity 1,000 mL Integrated polyester screen type Surface area 130 cm2 Pore size 32mm Polyester screen type Pore size 47mm
Polyurethane Sorin Kids neonate set, custom made, (Sorin Group, Mirandola, Italy)
Diameter1=4inch, length
1.10 m, 0.02 m2 contact surface area, P.h.i.s.i.o. coating
Sorin Kids neonate set, custom made (Sorin) Diameter1=4inch, length
2.95 m, 0.069 m2contact surface area,
P.h.i.s.i.o. coating
230 mL
Infant roller Sorin Kids D101 (Sorin) Hollow fiber Polycarbonate housing, polypropylene membrane 0.61 m2 Priming volume 87 mL P.h.i.s.i.o. coating Open hardshell, polycarbonate Minimum capacity 30 mL Maximum capacity 1,500 mL Sorin Kids D131 stand-alone arterial filter Polycarbonate housing, phosphoryl-chloride screen type membrane Surface area 27 cm2
Pore size 40mm Priming volume 28 mL
Polyester Pore size 51mm
Polyurethane Sorin Kids, custom made Diameter1=4inch, length
1.05 m, 0.02 m2contact surface area
P.h.i.s.i.o. coating
Sorin Kids neonate set, custom made Arterial part diameter1=4
inch, length 1.88 m Venous part diameter3/
8 inch, length 1.51 m Total 0.08 m2contact surface area P.h.i.s.i.o. coating 420 mL Pediatric centrifugal Revolution (Sorin) Pump casing polycarbonate Priming volume 57 mL Capiox FX15 (Terumo Europe) Hollow fiber Polycarbonate housing, polypropylene membrane 1.5 m2 Priming volume 144 mL X-coating Open hardshell polycarbonate Minimum capacity 70 or 200 mL Maximum capacity 3,000 or 4,000 mL Integrated polyester screen type Surface area 360 cm2 Pore size 32mm Polyester screen type Pore size 47mm
Polyurethane None Sorin Kids Pediatric set, custom made Diameter3/8inch, length
4.87 m, 0.15 m2contact surface area P.h.i.s.i.o. coating 700 mL
ARTICLE
IN
PRESS
A. van Saet et al. / Journal of Cardiothoracic and Vascular Anesthesia 00 (2019) 1 9 3Samples
Four-milliliter blood samples were taken from the arterial (post oxygenator) side of the circuit via a manifold sample port in a polypropylene (PLP) ethylenediaminetetraacetate tube (7.2 mg) (BD Vacutainer, BD Life Sciences, Plymouth, UK). Samples were taken immediately after injection of the drugs (T1) and after 2, 5, 7, 10, 30, 60, 180, and 300 minutes.
Samples were stored at 4˚C until processing. After centrifu-gation (10 min at 3,600 rpm), the supernatant serum was trans-ferred to PLP cryogenic vials with PLP screw caps (Sarstedt Aktiengesellschaft & Co, N€umbrecht, Germany) and stored at 80˚C until analysis.
Assay Methods
Drug concentrations for sufentanil, midazolam, propofol, and methylprednisolone were measured using liquid chroma-tography mass spectrometry. Methods were validated accord-ing to US Food and Drug Administration guidelines for bioanalytical method validation.10All analyses included qual-ity control samples, as is required for Food and Drug Adminis-tration analyses, and were performed in International Organization for Standardization- and Good Clinical Practice-certified laboratories by a Practice-certified research technician.
Drug concentrations for sufentanil were measured using a Thermo TSQ Vantage triple-stage quadrupole mass spectrome-ter (Thermo Fisher Scientific, Waltham, MA) at the pharmacy laboratory of the Erasmus Medical Center. Drug concentrations for midazolam were measured using a Quattro Premier mass spectrometer (Waters Corp, Milford, MA) at the pharmacy labo-ratory of the Erasmus Medical Center. Propofol was measured using a Thermo TSQ Quantiva triple-stage quadrupole mass spectrometer (Thermo Fisher Scientific) at the pharmacy labora-tory of the University Medical Center in Groningen, the Nether-lands. Drug concentrations for methylprednisolone were measured using a SCIEX Triplequad 6500+ mass spectrometer (AB SCIEX, Concord, Ontario, Canada) and Analyst software, Version 1.7 (AB SCIEX) at the Institute for Biomedical and Pharmaceutical Research in N€urnberg-Heroldsberg, Germany.
The lower limit of quantification was 0.25
m
g/L for sufenta-nil, 2.0m
g/L for midazolam, 100.0m
g/L for propofol, and 10m
g/L for methylprednisolone. The upper limit of quantifica-tion (ULOQ) was 50.0m
g/L for sufentanil, 2,400m
g/L for midazolam, unknown for propofol, and 30100m
g/L for meth-ylprednisolone.Statistical Analysis
The relationship between log-transformed drug concentra-tion (dependent variable) and type of CPB system and sample time point (independent variables) was assessed with linear mixed model analyses. Linear mixed model analyses were used because a correlation can be expected between repeated measurements of the same variable (ie, drug concentration) in the same subject (ie, individual CPB systems). Both indepen-dent variables were treated as categorical variables, and a 2-way interaction effect between the type of CPB system and sample time point was included in the model. To correct for within-system correlations between time points, a random intercept was used and we assumed a first-order autoregressive error covariance matrix. This model specification was chosen by comparing values of the Akaike information criterion between different structures for the random effects and the error covariance matrix.
For each time point and each CPB system, the difference between the predicted log-transformed estimated marginal means11at T1 and the predicted log-transformed concentration at the different time points was calculated, as was the 95% con-fidence interval (CI) of this difference. Finally, this difference and the 95% CI were exponentiated to obtain the percentage drug recovery (the percentage of drug still present in the prim-ing fluid) for T2 to T300. The maximum expected concentration in case of perfect mixture of drug (MEC) was calculated by dividing the amount of drug added to the CPB systems by the total priming volume used because it was unclear whether mix-ing of drug with the primmix-ing fluid would be complete at T1.
Spearman correlations were calculated to assess the rela-tionship between drug recovery at 60 and 180 minutes and log P, protein binding, and pKa among the 4 drugs (ie, with a sam-ple size of n = 4 drugs). For each drug, the recovery used for the calculation of this correlation was based on the estimated marginal means of the linear mixed model.
Statistical analyses were performed using SPSS Statistics for Windows, Version 24 (IBM Corp, Armonk, NY). All statis-tical tests used a 2-sided significance level of 0.05.
Results
No technical problems were encountered during the experi-ments. A total of 81 samples (27 for each CPB system cate-gory) were analyzed. No loss of drug samples occurred.
Fig 1shows predicted drug recovery (based on the estimated marginal means of the linear mixed models) versus time for propofol for each CPB system category. There was a sharp and significant decline in recovery of propofol compared with T1 in all the system categories in the first 60 minutes, to 41.3%
Table 2
Priming Fluid Composition
Neonatal Infant Pediatric Priming volume (mL) 263 430 683 RBC (mL) 135 235 365 FFP (mL) 30 40 50 Gelofusine (mL) 30 40 50 Albumin 20% (mL) 40 50 100 Mannitol 15% (mL) 20 50 100 NAHCO3 8.4% (mL) 8 15 18 Heparin (mL) 0.4 0.5 1 Flow (L/min) 0.5 1.5 3.5 Temperature (˚C) 36 36 36 Line pressure (mmHg) 100 100 100 Abbreviations: FFP, fresh frozen plasma; NAHCO3, Sodium bicarbonate; RBC, red blood cells.
ARTICLE IN PRESS
(95% CI 15.5-67.2), meaning that approximately 59% of the added drug was lost from the circulating prime fluid at that time. After that, recovery continued to decrease slightly but significantly throughout the study period to 25% (95% CI 4.7-45.3) after 180 minutes and 19% (95% CI 5.2-32.8) after 300 minutes. Based on the interaction effects in the mixed models, there was a significant difference in the pattern of recovery over time among systems for propofol (p< 0.001). Recovery was the greatest in the infant system, followed by the pediatric system. The neonatal systems appear to absorb the largest amount of propofol to their system components.
Fig 2shows predicted drug recovery (based on the estimated marginal means of the linear mixed models) versus time for sufentanil, midazolam, and methylprednisolone. Because there was no significant interaction effect between the type of CPB system and sample time point for sufentanil (p = 0.111), mida-zolam (p = 0.213), or methylprednisolone (p = 0.829), a single graph was used to depict decrease of drug recovery over time for all 3 systems. The pattern of decline in recovery of sufenta-nil shows stable drug concentrations in the first 7 minutes, with a significant decline compared with T1 from T10 onward. Drug recovery was 41.7% (95% CI 35.9-47.4) at 60 minutes. After 60 minutes, recovery continued to decrease slightly but significantly throughout the study period. For sufentanil, recovery was 23.0% (95% CI 9.2-36.8) after 180 minutes and 15% (95% CI 1.3-31.3) after 300 minutes.
For midazolam, there also was stable drug recovery in the first 7 minutes. The decline in drug recovery compared with T1 reached significance at T10 and from T60 forward. Drug recovery was 87.3% (95% CI 64.9-109.7) after 60 minutes and 82.0% (95% CI 64.6-99.4) after 180 minutes. For methylpred-nisolone, there was stable recovery of drug in the first
30 minutes. After that, recovery increased significantly com-pared with T1 to values much higher than 100%.
No significant correlation between log P and percentage recovery of the 4 drugs at 180 minutes (
r
0.324; p = 0.304) was found. In addition, a decreased recovery of highly protein-bound drugs (r
=0.822; p = 0.007) was found. The third fac-tor that correlated to percentage recovery at 180 minutes in the present study was pKa (r
= 0.822; p = 0.007).Discussion
These in vitro experiments investigated drug recovery in 3 different pediatric CPB systems used in the authors’ center. Decrease of drug concentration in the circulating prime fluid for propofol and sufentanil was fast in the first 60 minutes. Because 60 minutes is a relatively common bypass time in pediatric cardiac surgery, this period is clinically very relevant. The decreasing speed of reduction in drug concentration after 60 minutes may be an indication of near complete saturation of binding places on the different components of the CPB sys-tems. It is, however, unknown whether there is a finite amount of binding places and if complete saturation of these binding places is possible. Hammaren et al.12and Myers et al.13 have shown that for propofol there appears to be no maximum satu-ration of binding places in complete adult CPB systems, even at very high propofol concentrations. Binding of propofol may be concentration dependent.12In contrast, complete saturation of oxygenator membrane fragments has been shown for fenta-nyl in in vitro studies.5
In the present study, midazolam recovery was remarkably large in both centrifugal and roller-pump systems. Unfortu-nately, there are no in vitro studies of pediatric CPB systems with which to compare the present study’s results. An ECMO study performed in the authors’ hospital showed a recovery pattern similar to the present experiment in systems with a cen-trifugal pump.8 In roller-pump systems there was just 7.5% recovery after 2 minutes and 0.6% recovery after 180 minutes. For midazolam, however, it must be taken into account that the concentrations measured in the present study were far above the ULOQ. The authors believe that this may at least partly be the reason that the MEC was lower than the measured concentrations. This introduced an unknown amount of bias, but the authors do not expect that this measurement bias would explain the high recovery rates. An error in the addition of medication to the system or a laboratory error also were not expected because the experiments were performed on different days for the different systems and all the percentage recovery versus time curves show a similar pattern.
Methylprednisolone concentrations in the first hour were much lower than would be anticipated from the MEC. This most likely was caused by a problem with the mixing of meth-ylprednisolone added to the system with the priming fluid. Another explanation would be very fast binding of methyl-prednisolone to components of the CPB system and release of drug from those binding sites after 60 minutes. However, it is unknown whether binding of drug to components of a CPB system is a reversible process.
Fig 1. Drug recovery versus time based on estimated marginal means for pro-pofol, for each cardiopulmonary bypass system category, expressed as means and 95% confidence interval based on the linear mixed models.
MEC, maximum expected concentration in case of perfect mixture of drug. *p< 0.002.
ARTICLE IN PRESS
The substantial differences in recovery found between drugs suggest that drug characteristics influence the interaction with components of the CPB system (Table 3). From previous studies in ECMO systems, recovery of drugs seems to be highly depen-dent on lipophilicity.8,9,14,15In the present study, no correlation between log P and recovery percentage was found. This likely was caused by the recovery profile of midazolam. Significantly decreased recovery of highly protein-bound (>80%) drugs also was shown in a previous publication.15 We found a generally decreased recovery of highly protein-bound drugs, which may be caused by binding of drug to protein adhered to system com-ponents.15The similar recovery patterns of propofol and sufen-tanil suggest that there is a common physicochemical property
of both drugs causing this effect. The authors, however, cannot explain why the recovery pattern of midazolam is different in our study because the physicochemical properties known to influence recovery are very similar to those of propofol and sufentanil. Another factor correlated to the percentage recovery at 180 minutes in the present study was pKa. To the authors’ knowledge no correlation between pKa and recovery has been described previously. The surface-coated CPB systems are neg-atively charged, making electrostatic attraction of positively charged molecules a possible mechanism for absorption of drugs to CPB system components. Drugs with a high pKa are unlikely to be dissociated at normal pH, however, which was maintained during the study period.
Fig 2. Drug recovery versus time based on estimated marginal means for midazolam, sufentanil, and methylprednisolone, expressed as means and 95% confidence interval based on the linear mixed models.
MEC, maximum expected concentration in case of perfect mixture of drug. *p< 0.05
ARTICLE IN PRESS
A significant difference in drug recovery between the differ-ent types of CPB system was found only for propofol. In gen-eral, one would expect a larger system to have more binding sites for drugs and thus a lower recovery for drugs with similar properties. The surface area of the oxygenator and the PVC tubing in the authors’ pediatric systems are much larger than those in their neonatal and infant systems. In the present study, midazolam and methylprednisolone showed lower recovery in larger systems, although not significantly so. For propofol, however, there was greater recovery in larger systems. The dif-ferences are smaller than would be expected if system size were the only factor involved.
Different components of the CPB system are capable of absorbing drugs to their plastics. Differences in drug recovery between different types of oxygenator have been described extensively.5,16 With the new polymethylpentene and PLP membranes the oxygenator does not appear to be a factor of considerable interest in drug recovery anymore. Based on a study by Preston et al., 80% of drug is lost to PVC tubing, with a small additional amount of drug (of just 5%) lost to the oxy-genator.17 Silicone tubing has been shown to decrease the recovery of drugs compared with PVC tubing.2,8The effect of different surface coatings on both oxygenator and tubing on drug recovery has been investigated by several authors,7,12,13 and those studies suggest different effects for different coat-ings for different types of drugs. The addition of an arterial fil-ter also may lead to decreased drug recovery.13Many factors thus are at play and interact with each other.
It is unclear which differences in system composition play a role in the present study. The interplay of differences in sur-face area, coating, tubing type, and pump type makes it diffi-cult to draw firm conclusions about the influence of individual system components on drug recovery. In an earlier study by Preston et al., a Terumo Baby Rx oxygenator was used17. The Terumo Capiox Fx05 was used in the present study’s neonatal systems, which is the same oxygenator with the same mem-brane, the same coating, and the same surface area, but with an integrated arterial filter. The Baby Rx oxygenator absorbed 3% of fentanyl added to the system in the study by Preston et al., which amounts to 0.6 ng/cm2. Raffaeli et al. showed that sufentanil absorption is similar to fentanyl absorption in their ECMO systems.9Assuming the aforementioned holds true, the amount of drug absorbed in the present study’s neonatal sys-tems just by the tubing and the arterial filter would be 4.2
m
g of sufentanil at 180 minutes (total uptake of sufentanil 72% at 180 min of 10m
g added to the neonatal systems minus theamount absorbed by the oxygenator). Because the neonatal systems in the present study had silicone tubing, this relatively low amount of absorption seems unlikely.
A similar calculation for the pediatric systems used in the present study is possible. The Capiox Fx15 oxygenator would absorb 9
m
g in total because of its larger surface area. After 180 minutes, 76% of the 60m
g of sufentanil added to the sys-tem would be absorbed. Further calculation shows that this would mean that the Sorin tubing would absorb 25.1 ng/cm2of sufentanil. This, however, would mean that all the sufentanil in the neonatal system in the present study would have to have been absorbed, which is not the case. For a more extensive cal-culation, the reader is referred to the supplemental materials.Hynynen et al.18described propofol recovery of 25% after 120 minutes of circulation in an adult system. Hammaren et al.12and Myers et al.13described recovery of 37% and 43%, respectively, after 60 minutes. These values are remarkably similar to those of the present study, even though completely different CPB systems were used. Based on the calculations previously described and the results by Hynynen, Hammaren, and Meyers, it appears that it is very difficult to translate research performed in individual centers to one’s own clinical practice because of the amount of factors and interactions at play. It is clear that not all factors influencing absorption to dif-ferent components of CPB systems are known. A possible lack of generalizability thus may be seen as a limitation to the pres-ent study and other studies already performed in this field.
Several authors have found no influence of temperature management on the recovery of drugs in their systems.16,19,20 Therefore the authors of the present study did not attempt to simulate a cooling protocol.
Despite the significant decrease in recovery of drug from the priming fluid of the CPB system found in the authors’ in vitro studies, clinically patients do not wake up on initiation or dur-ing CPB. A clinical study of propofol infusions in cardiac sur-gery in adult patients showed no change or even a decrease in bispectral index values during CPB.21 This most likely was caused by an increase in unbound drug concentration as a result of decreased protein concentration on initiation of CPB,21-23causing a greater amount of drug available for end-organ effect.
The present study has several other limitations. Only com-plete systems were tested, thus there is no information in the present study as to which amount of drug was absorbed by which system component. Unfortunately this type of research is expensive because of the costs of systems, blood products for
Table 3
Drug Physicochemical Data19
Blood/Plasma Ratio Log P Protein Binding (%) Vd (L/kg) pKa Methylprednisolone Unavailable 2.06 78 Unavailable 12.58 Sufentanil 0.75 3.4 Unavailable Unavailable 8.86 Propofol Unavailable 3.81 95-99 Unavailable for children 10.98 Midazolam 0.75 3.89 97 Children 6 mo-16 y 1.24-2.02 10.98 Abbreviations: Vd, volume of distribution.
ARTICLE IN PRESS
priming, and drug concentration measurements. Also, for pur-poses of the present study, namely the integration of these in vitro results with the results of the in vivo part of the CPB PHARM study, the authors sought to mimic everyday practice in their hospital as closely as possible, making testing of indi-vidual components less useful. For the same reasons, the authors have not performed isolated drug studies. It is not known whether there is competition for binding sites to components of the CPB system for different individual drugs. Although theo-retically this is an interesting topic, in clinical practice, patients receive multiple medications at the same time.
Just 3 full systems in each category were used, and although this may seem like a small number, the sample size is compa-rable with that of other publications performed with both CPB and ECMO systems and which are cited in this article.
Spontaneous degradation may have produced bias in the pres-ent study because it causes a decrease in drug concpres-entrations not caused by adherence of drug to components of the CPB system. Previous studies in the authors’ hospital have shown that sponta-neous degradation over 24 hours for midazolam is 11.4% and for sufentanil is 0%.9For propofol, spontaneous degradation in glass bottles with daylight and room temperature is around 5% after 6 hours.24 For methylprednisolone, no references were found for spontaneous degradation. Because degradation usu-ally is calculated over 24 hours, the effect on the present study’s results would be constant over time.
For the calculation of MEC, the authors did not correct for blood-plasma ratio. In previous studies in the authors’ hospital, midazolam was shown to have a blood-plasma ratio of 75%.8 Because the present study’s MEC value was calculated in blood, but drug concentrations were measured in plasma, MEC would be underestimated for midazolam. Because pro-pofol is highly bound to red blood cells in vivo, a high blood-plasma ratio would be expected and the present study’s MEC would be overestimated.
The authors did not aim for a similar drug concentration in the different types of CPB systems. Instead, the dose of drug that would be administered to a typical patient connected to a neonatal, infant, or pediatric CPB system in the authors’ daily clinical setting was added because the goal of the pres-ent study was to mimic the authors’ everyday practice as closely as possible so that in the future the authors will be able to incorporate CPB system recoveries in a larger popu-lation pharmacokinetic model of the influence of CPB on drug concentrations in children.
Because of the high doses of drug added to the authors’ sys-tems, drug concentrations were more than the ULOQ for some drug assays, necessitating additional dilution before quantifica-tion. The high doses also might have resulted in potential dif-ferences in absorption rates, as has been shown for propofol.12
In the present study, indications that mixing of drug with the priming fluid is not always complete were observed. This may have clinical consequences if drugs are added to the CPB sys-tem during CPB, rather than given directly to the patient.
Despite these limitations, to the authors’ knowledge, the present study is the first comprehensive in vitro testing of CPB systems used in pediatric congenital cardiac surgery.
In conclusion, the present study’s in vitro experiment with neonatal, infant, and pediatric CPB systems shows a variable recovery of routinely used drugs with significant differences among drugs but not among system categories, except for pro-pofol. The study also demonstrates that the generalizability of this type of research may be limited. The clinical consequen-ces of the present study’s research must be investigated fur-ther. The decreased recovery of sufentanil and propofol could lead to suboptimal dosing of patients during cardiac surgery with the use of CPB, even though clinically this doesn’t show; thus it is important that these findings are integrated with the results of in vivo studies into population pharmacokinetic models to further investigate the clinical relevance of the pres-ent study’s findings and the implications for perioperative patient care.
Conflict of Interest
The authors declare no conflicts of interest. Acknowledgements
The authors thank Antony van Dijk, clinical perfusionist, Department for Cardiothoracic Surgery, Erasmus Medical Center, Rotterdam, The Netherlands, for his help in the study design.
Supplementary materials
Supplementary material associated with this article can be found in the online version at doi:10.1053/j.jvca.2019.08.029. References
1van Saet A, de Wildt SN, Knibbe CA, et al. The effect of adult and pediat-ric cardiopulmonary bypass on pharmacokinetic and pharmacodynamic parameters. Curr Clin Pharmacol 2013;8:297–318.
2Koren G, Crean P, Klein J, et al. Sequestration of fentanyl by the cardio-pulmonary bypass (CPBP). Eur J Clin Pharmacol 1984;27:51–6.
3Rosen D, Rosen K, Davidson B, et al. Fentanyl uptake by the scimed mem-brane oxygenator. J Cardiothorac Anesth 1988;2:619–26.
4Durkan W, Lonergan M, Schwartz S, et al. Effect of membrane oxygenator on sufentanil blood levels during cardiopulmonary bypass. Anesth Analg 1988;67:54.
5Rosen DA, Rosen KR, Silvasi DL. In vitro variability in fentanyl absorp-tion by different membrane oxygenators. J Cardiothorac Anesth 1990;4:332–5.
6Hynynen M. Binding of fentanyl and alfentanil to the extracorporeal cir-cuit. Acta Anaesthesiol Scand 1987;31:706–10.
7Preston TJ, Ratliff TM, Gomez D, et al. Modified surface coatings and their effect on drug adsorption within the extracorporeal life support cir-cuit. J Extra Corpor Technol 2010;42:199–202.
8Wildschut ED, Ahsman MJ, Allegaert K, et al. Determinants of drug absorp-tion in different ECMO circuits. Intensive Care Med 2010;36:2109–16.
9Raffaeli G, Allegaert K, Koch B, et al. In vitro adsorption of analgoseda-tive drugs in new extracorporeal membrane oxygenation circuits. Pediatr Crit Care Med 2018;19:e251–8.
10 Food and Drug Administration. Guidance for industry bioanalystical method validation. Available at: https://www.fda.gov/files/drugs/pub-lished/Bioanalytical-Method-Validation-Guidance-for-Industry.pdf. Accessed September 5, 2019.
ARTICLE IN PRESS
11 Deng DC: Least squares means (marginal means) vs. means 2009. Avail-able at: http://onbiostatistics.blogspot.com/2009/04/least-squares-means-marginal-means-vs.html.
12Hammaren E, Rosenberg PH, Hynynen M. Coating of extracorporeal cir-cuit with heparin does not prevent sequestration of propofol in vitro. Br J Anaesth 1999;82:38–40.
13Myers GJ, Voorhees C, Eke B, et al. The effect of Diprivan (propofol) on phosphorylcholine surfaces during cardiopulmonary bypass—an in vitro investigation. Perfusion 2009;24:349–55.
14Shekar K, Roberts JA, Barnett AG, et al. Can physicochemical properties of antimicrobials be used to predict their pharmacokinetics during extra-corporeal membrane oxygenation? Illustrative data from ovine models. Crit Care (London, England) 2015;19:437.
15Shekar K, Roberts JA, McDonald CI, et al. Protein-bound drugs are prone to sequestration in the extracorporeal membrane oxygenation circuit: Results from an ex vivo study. Crit Care (London, England) 2015;19:164.
16Booth BP, Henderson M, Milne B, et al. Sequestration of glyceryl trinitrate (nitroglycerin) by cardiopulmonary bypass oxygenators. Anesth Analg 1991;72:493–7.
17Preston TJ, Hodge AB, Riley JB, et al. In vitro drug adsorption and plasma free hemoglobin levels associated with hollow fiber oxygenators in the
extracorporeal life support (ECLS) circuit. J Extra Corpor Technol 2007;39:234–7.
18Hynynen M, Hammaren E, Rosenberg PH. Propofol sequestration within the extracorporeal circuit. Can J Anaesth 1994;41:583–8.
19Skacel M, Knott C, Reynolds F, et al. Extracorporeal circuit sequestration of fentanyl and alfentanil. Br J Anaesth 1986;58:947–9.
20Hickey SQI, Gaylor JD. In vitro variability in propofol absorption by dif-ferent membrane oxygenators. Crit Care 2001;5(Suppl C):8.
21Takizawa E, Hiraoka H, Takizawa D, et al. Changes in the effect of propo-fol in response to altered plasma protein binding during normothermic car-diopulmonary bypass. Br J Anaesth 2006;96:179–85.
22Jeleazcov C, Saari TI, Ihmsen H, et al. Changes in total and unbound concentrations of sufentanil during target controlled infusion for car-diac surgery with cardiopulmonary bypass. Br J Anaesth 2012;109: 698–706.
23Dawson PJ, Bjorksten AR, Blake DW, et al. The effects of cardiopulmo-nary bypass on total and unbound plasma concentrations of propofol and midazolam. J Cardiothorac Vasc Anesth 1997;11:556–61.
24Sautou-Miranda S LE, Groueix M, Chopineau J. Compatibility of propofol diluted in 5% glucose with glass and plastics (polypropylene, polyvinyl-chloride) containers. Int J Pharmaceutics 1996;130:251–5.
